Dear ARM team,
You guys are absolute legends. The Cortex-M series is still the undisputed king 15+ years later. M3/M4/M7 are basically perfect — performance, power efficiency, tools, ecosystem — everything just clicks. The entire embedded community loves you for that.
But let’s be honest: since the M33/M7 era there hasn’t been a truly revolutionary core that became widely available to ordinary developers. Newer stuff like M55 and M85 looks amazing on paper, but for makers, small companies and hobbyists they remain locked behind massive NRE fees or exclusive partner programs. Most of us are still stuck on M3/M4 simply because that’s the only thing we can actually use legally and easily.
Meanwhile the FPGA world has moved on. People are already cramming almost a complete STM32F103 (full Cortex-M3 + advanced TIM1 + GPIO ports + NVIC + SysTick) into a $20 Tang Nano 20K board and still have thousands of LUTs left. Open-source repos doing exactly that are all over GitHub right now.
And here’s the insane opportunity nobody is talking about.
Imagine ARM simply added a new section on developer.arm.com called Configure & Buy Your Cortex-M Core.
You go to the page, pick the base core (M0, M3, M4, M7, M33, M55 — whatever is possible), then a web configurator (think CubeMX but for the processor itself) lets you strip everything you don’t need: drop the MPU, FPU, DSP extensions, TrustZone, SAU, ETM, ITM, DWT, bit-banding, TCM interfaces, optional instruction set variants — literally everything that isn’t required for your project.
The page instantly shows you the resulting size in LUTs for Xilinx, Intel, Lattice and Gowin, maximum achievable frequency, power estimate, and most importantly — the price. A bare-bones M3 for $10, a fully loaded M55 for $49.
You pay, and immediately download encrypted RTL or even a pre-synthesized netlist targeted to your chosen FPGA family, plus a royalty-free license (commercial use included) and one year of free updates.
This would be a literal money-printing machine with almost zero marginal cost.
Right now ARM makes money two ways: big upfront licenses to giant corporations (awesome) and per-chip royalties (also awesome).
This would add a third revenue stream — micro-transactions from hundreds of thousands of hobbyists, indie developers, tiny startups, universities and research labs all over the planet. Ten thousand people spending $25 a year is already a quarter million dollars. A hundred thousand is $2.5 million. A million is real money. And the real number would be way higher.
The upside for ARM is enormous: new revenue stream built once and then running forever, insane community goodwill, people stop fleeing to RISC-V out of desperation, every student on Earth learns ARM instead of the competition, drastically reduced piracy of DesignStart cores.
The upside for us is finally being able to build truly minimal, rock-solid, errata-free controllers. No unused USB stack, no ADC you’ll never touch, no hidden bugs in features you disabled anyway. Safety-critical drone flight controllers, medical devices, industrial equipment — all running on a $30 FPGA board with exactly the core you want and nothing else.
Nobody else is offering this today. SiFive kind of tries, but it’s slow and expensive. ARM could own this market overnight with one simple web tool.
Please, seriously think about it. We don’t want to leave the ARM ecosystem — we love it and want to stay forever. Just give us a way to keep giving you money in a form that actually fits small teams and individual engineers.
Make buying a Cortex core as easy and impulse-friendly as buying a Tang Nano or an ESP32.
With massive respect and genuine love from a community that has been living and breathing your cores for over fifteen years.
— an embedded developer who really, really wants to keep paying ARM